Blog

Notes on teaching robots new tricks

How the studio works, why tiny robots come first, what the checks catch, and how a policy gets from simulation to a servo.

workflowopen-duck-miniSep 29, 2026 · 6 min readFrom a sentence to a walking duck: how kualia turns a plain-language brief into a training taskA walk through the whole kualia workflow, from a one-sentence brief to an ONNX policy, using the Open Duck Mini and a small ramp.Read the post

Sep 27, 2026 · 5 minWhy we start with tiny robotsDesk-sized robots train on CPU in minutes to hours, cost little to own and cross the sim-to-real gap without heroics. That is why kualia's shelf starts with them.

Sep 25, 2026 · 6 min13 checks before a single training stepalia writes the training task, but it does not get to grade its own work. Every task passes 13 sandbox checks before training can start. Here is what each one catches.

Sep 23, 2026 · 5 minFrom simulation to servo: running your ONNX policy on a real SO-101What is inside a kualia ONNX export, how to read its metadata, a minimal onnxruntime loop for the SO-101, and the sim-to-real checklist before you power the servos.

Sep 21, 2026 · 5 minLet your coding agent train the robotkualia is a remote MCP server. Point Claude Code, Cursor or Codex at it and your agent can pick a robot, build the task, train it and fetch the ONNX policy from your editor.

Research digest

Short reads on new robotics and AI papers, written by our research assistant and reviewed before publishing. 1,060 so far.

Page 13 of 59

May 15, 2026The Generative Cascade: Synthetic Data and the Recursive Training of Large Language ModelsThe alignment of large language models with human intent has largely depended upon access to high quality, diverse instruction following data. While early efforts relied on expensive human annotation, the Self-Instruct p May 15, 2026Cross-Architecture Distillation for Diffusion Language Models: A Technical Analysis of TIDEDiffusion large language models have emerged as a compelling alternative to autoregressive architectures, offering the theoretical advantages of parallel decoding and bidirectional context modeling. Yet despite these arc May 15, 2026Neural Harmonic Textures: Reconciling Primitive Efficiency with Neural ExpressivenessThe divide between explicit geometric primitives and implicit neural representations has defined the recent evolution of novel view synthesis. Since 3D Gaussian Splatting displaced Neural Radiance Fields as the dominant May 15, 2026Bridging the Modality Divide: Universal Skeleton Understanding Through Differentiable RenderingThe proliferation of Multimodal Large Language Models has created a paradox. While MLLMs demonstrate remarkable reasoning capabilities across visual and linguistic domains, they remain fundamentally constrained by their May 15, 2026Embedded Language Flows: Reconciling Continuous Diffusion with Discrete Text Introduction The application of diffusion models to language generation has long presented a fundamental architectural tension. While diffusion and flow based approaches have become dominant for continuous data such as May 15, 2026The Hidden Economics of AI Agents: Why Your Coding Assistant Costs More Than You ThinkAs software engineering workflows increasingly integrate autonomous AI systems, a quiet economic crisis is emerging. While early large language models operated as simple chat interfaces with predictable per query costs, May 15, 2026Rethinking Supervision in Novel View Synthesis: How Monocular Depth Enables Internet-Scale Training Introduction The human visual system possesses a remarkable capacity for spatial reasoning. Presented with a single photograph of a cathedral, we effortlessly extrapolate the geometry of flying buttresses, infer occlude May 15, 2026 May 15, 2026Transferable Physics-Informed Representations: Closing the Gap Between Learning and Inference in Scientific Machine Learning Introduction Physics-informed neural networks (PINNs) have emerged as a compelling bridge between deep learning and computational physics, offering a mesh-free alternative to traditional numerical methods for solving pa May 14, 2026Streaming Adaptation in Production ASR Systems: A Technical Analysis of Online Continual LearningThe deployment of automatic speech recognition systems in production environments presents a fundamental tension between model stability and adaptability. While research on continual learning has traditionally focused on May 14, 2026Generalization Through Chaos: How Neural Networks Learn at the Edge of Stability Introduction The optimization of deep neural networks presents a persistent theoretical puzzle. Classical convex optimization dictates that gradient descent remains stable only when the learning rate satisfies $\eta < 2 May 14, 2026From Static Geometry to Functional Mechanisms: The Challenge of Dynamic CAD in Agentic Systems Introduction The past year has witnessed a notable shift in how we approach computational design. Rather than training monolithic generative models on vast datasets of 3D shapes, researchers have increasingly turned to May 14, 2026Bridging Local and Global Temporal Dynamics in Brain-Computer Interfaces: A Critical Look at EEG-MFTNetBrain-computer interfaces represent one of the most promising yet technically demanding frontiers in neurotechnology. While the vision of controlling prosthetic limbs or wheelchairs through thought alone has captivated r May 14, 2026cuGenOpt: Bridging the Trilemma of Generality, Performance, and Usability in GPU-Accelerated OptimizationCombinatorial optimization problems form the computational backbone of modern logistics, manufacturing, and resource allocation. Yet practitioners face a persistent trilemma when selecting solution methods. Exact approac May 14, 2026Self-Supervised GANs for Electron Microscopy: Rethinking Model Complexity in Scientific ImagingElectron microscopy stands as one of the most powerful tools for interrogating material structure at the atomic scale, yet the translation of raw imaging data into scientific insight remains bottlenecked by the painstaki May 14, 2026 May 14, 2026 May 14, 2026The Mechanics of Steering: How Refusal Vectors Actually Reshape Language Model BehaviorSteering vectors have emerged as a lightweight alternative to fine-tuning for aligning large language models. By adding a learned direction to hidden activations at specific layers during inference, practitioners can ind